Repository navigation
__pow__(float, float) should be float | complex, not Any #12895
Description
Exponentiating a float with fractional power can be float or a complex, but nothing else:
In [1]: 1.0 ** 0.5
Out[1]: 1.0
In [2]: (-1.0) ** 0.5
Out[2]: (6.123233995736766e-17+1j)
But the result is currently typed as Any, but float | complex would be stricter and more self-explanatory
I don't see any technical reason for not doing so, maybe I'm missing something? But getting Any and then have mypy --strict complain about it is definitely a headscratcher.
Related issues:
Activity
AlexWaygood commentedon Oct 24, 2024
We've tried this before, and it causes too many false positive errors unfortunately. Most people most of the time are only using
**between two positive floats, so it's very confusing for those users to find that mypy infers the result asfloat | complexwhen they know it's afloatat runtime. You can see some discussion in #7682.Akuli commentedon Oct 24, 2024
Perhaps we could use
float | Any.hauntsaninja commentedon Oct 24, 2024
See #7682 (comment) , int | Any could be worth a shot
embray commentedon Oct 29, 2024
I don't think we should be doing something wrong just because some users didn't learn basic maths